Yazelix without nix

[abrassel]

Hi folks, how easy would it be to get yazelix working without installing nix?

[luccahuguet]

Hi there

You'd kind of lose pack functionality and have a long list of things to install and lose the ability to switch features on/off

But it should not be hard

What are you trying to get out of it? So i can understand you better

[luccahuguet]

Good to hear yazelix fits you well

I'll create an opinionated slim profile of yazelix using rust only tools

And use cargo bin and cargo update to install them

This should fix your problem

[abrassel]

That would be amazing! I hope other folks can benefit from your hard work, too. I know that I will :) Cheers!

[luccahuguet]

That would be amazing! I hope other folks can benefit from your hard work, too. I know that I will :) Cheers!

Surely others will too

and devenv is a bit heavy too, so it's good to have alternatives

Cheers!

[petertheprocess]

Hi, for me the disk space is a bottleneck, nix and devenv us too heavy and I already installed yazi, helix and zellij. Maybe it is better to provide a script or guide to allow us put those nushell scripts and config files in the right place?

[luccahuguet]

hi @petertheprocess , I'll investigate this soon and get back to you

possibly disable_nix = true and/or disable_devenv

if it helps some users and does not add too much complexity, it's worth it

[luccahuguet]

Related @abrassel

DeterminateSystems/nix-installer#214

[jigypeper]

I think it might be worth having a branch that is "nix-less", maybe based on an earlier version? (like version 8 or 9)

[luccahuguet]

It's my next priority (on main, though)

Also, Ive been thinking about adding a roadmap section to the readme (or a discussion)...

But the roadmap should be something like

• nixless mode
• the yazelix website
• easy way to access yazelix.toml or a config menu
• create a full blown nixpkgs yazelix package
• create the yazelix zellij plugin
• maybe create a yazi plugin as well to improve the file opening flow
• rewrite some parts of the codebase in rust

[luccahuguet]

Hey everyone

I've created a nixless-system-mode branch that allows running yazelix with system-installed tools (zellij, yazi, helix/nvim, etc.) without requiring nix or devenv: https://github.com/luccahuguet/yazelix/tree/nixless-system-mode

it's basically v12 + nixless mode

However, this branch will not be actively maintained. I may update it once every 6 months to a year to sync with major changes, but no guarantees

yazelix's roadmap is heading deeper into the nix ecosystem — a nixpkgs package, zellij/yazi plugins, helix plugin support, AI tooling integrations, and eventually some rust rewrites

Every one of these features assumes a managed environment. Maintaining two parallel codepaths for every new feature isn't very practical for a one-person project

The nixless branch is a snapshot that likely works today (I havent had to time to test it yet)

If you can't or don't want to use nix, it's there for you to use, or fork, just know you're accepting the trade-off of potentially missing future features and fixes. I will gladly accept PR's to it, and help when possble, but with a lower priority

For those willing to give nix a shot: the disk space concern is valid, but devenv/nix provide reproducibility and feature toggling that's hard to replicate otherwise. It's an investment that pays off IMO

We do need a way to install nix without admin privileges, I'll look into it

Thanks for understanding

also, I really recommend Zed with helix mode (no nix needed)

when i do test it, i'll let you know

[jigypeper]

This is a good step forward, especially for people working in constrained environments.

If zed is a no go, try out Emacs with the Helheim configuration.

https://github.com/anuvyklack/helheim-emacs

It's a great computing environment by itself.

[luccahuguet]

the mise backend option is looking promissing, and would fix both disk usage and admin rights problem

[petertheprocess]

What about pixi https://prefix.dev/ ?

[luccahuguet]

very interesting

[luccahuguet]

we now have a flake and it's never been easier to install yazelix

• install nix with flakes enabled (we recommend determinate nix)
• nix run --refresh github:luccahuguet/yazelix#install
• optionally: yzx desktop install
• profit

so it's not nixless, but at least you're actually earning the nix advantages for real

[luccahuguet]

after the first install, you can just yzx update runtime to keep yazelix up to date

[luccahuguet]

the devil on your shoulder:

pay the price, it's worth it!

[luccahuguet]

Hey

I’m not planning to turn main Yazelix into a dual-backend product with a full non-Nix path beside the current managed one. That would simply add too much complexity

What I am actively moving toward is a cleaner architecture that separates package-management/backend concerns from the workspace UX layer. If that work lands the way I expect, it should make a lighter Yazelix Core edition genuinely feasible:

• keep the Zellij/Yazi/Helix UX and generated config layer
• drop managed package/environment ownership
• make dependency expectations explicit through a shared runtime contract/checker

So the direction here is real. I’m just not calling it a supported product yet before the boundary is actually proven in the codebase

The current planning/issues for this direction are:

• Separate environment-backend concerns from workspace UX and prepare a future Yazelix Core edition #531
• Define the runtime dependency contract and launch preflight checker scope #532
• Implement a shared runtime contract checker and reuse it across launch and doctor #533
• Extract a backend adapter boundary around env, run, refresh, and launch bootstrap #534
• Audit and classify yzx command families by backend coupling #535
• Define the Yazelix Core edition boundary and support story #536

Even if something like Yazelix Core lands, it will not have feature parity with full Yazelix, but it's ok, it will still be awesome, even though the managed environment is a big part of what makes Yazelix feel cohesive

And I do recognize the practical constraint behind this request:

• the storage cost of the full managed stack can be too steep
• not every user has admin privileges or the freedom to install Nix on their machine

For now, the closest related experiment is still:
https://github.com/luccahuguet/yazelix/tree/nixless-system-mode

The difference is that the current plan is to get there by improving the main architecture first, rather than by bolting on a second codepath that would be painful to maintain

Also, Im not promissing estimates, but we are probably getting there eventually

[jigypeper]

I've had to move over to Emacs now. Working in a restricted environment where I can't install tools, but found a good emacs config that has a good helix emulation layer. Helheim

[luccahuguet]

Ok, just an update

Yazelix no longer acts as a package manager
(Except that our home-manager module allows for great granular control over what gets installed)

And we lean a lot in nix and home-manager

There's no plan for any other way of installing yazelix currently

Nix is just that good, and with home-manager the control is even better

Using anything else means downgrade and implementation pain and higher complexity

[luccahuguet]

i think brew support is next